Analytics
4.9K members online now
Understand information in your reports and troubleshoot reporting issues such as self-referrals, (not set) data, and inaccurate information
 
Guide Me
star_border
Reply

Average page load time by specific browsers not real average - how is it calculated?

ACG
Follower ✭ ✭ ✭
# 1
ACG
Follower ✭ ✭ ✭

Dear all,

 

I have been asked to check the average page load time for our key browsers.

 

If I apply a filter to include only these browsers, I add up their averages and divide it by the total number of browsers I have, the average I get is different from the one summarised at the top of the table by Google Analytics but if I export the table into an Excel and select the relevant browsers, the average is correct. Why does this happen, please? 

average_from_summary_table.JPGaverage_from_excel.JPG

Also if I check all browsers' average page load time in my Excel the average I get is 2.14 and not 4.22 like GA says. How is this average calculated?

 

Any help on this would be really appreciated.

 

Thanks, A.

 

1 Expert replyverified_user
1 ACCEPTED SOLUTION

Accepted Solutions
Marked as Best Answer.
Solution
Accepted by topic author ACG
February 2017

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 2
Rising Star

Hi A,

 

In general you shouldn't average averages.  You would need the underlying raw data to do that.   If you have filtered for the browsers in question GA should handle that for you on the scorecard.

 

That said, you also should be aware that most sites that get less than 1MM hits per day have inadequate site speed samples.  This is also compounded by the fact that site speed sampling is client side and not all broswers comply.  

 

First thing you should check is the sample rate at which GA is collecting this data.  You are allowed up to 10k samples per day so adjust your rate so evenly distribute those throughout the day.  After that, let GA do the averaging for you.  Smiley Happy

 

https://developers.google.com/analytics/devguides/collection/analyticsjs/field-reference#siteSpeedSa...

 

 

Hope that helps!

 

Theo Bennett

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

View solution in original post

Marked as Best Answer.
Solution
Accepted by topic author ACG
February 2017

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 2
Rising Star

Hi A,

 

In general you shouldn't average averages.  You would need the underlying raw data to do that.   If you have filtered for the browsers in question GA should handle that for you on the scorecard.

 

That said, you also should be aware that most sites that get less than 1MM hits per day have inadequate site speed samples.  This is also compounded by the fact that site speed sampling is client side and not all broswers comply.  

 

First thing you should check is the sample rate at which GA is collecting this data.  You are allowed up to 10k samples per day so adjust your rate so evenly distribute those throughout the day.  After that, let GA do the averaging for you.  Smiley Happy

 

https://developers.google.com/analytics/devguides/collection/analyticsjs/field-reference#siteSpeedSa...

 

 

Hope that helps!

 

Theo Bennett

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Average page load time by specific browsers not real average - how is it calculated?

ACG
Follower ✭ ✭ ✭
# 3
ACG
Follower ✭ ✭ ✭

Hi Theo,

 

Thanks for your advise. I take note.

 

In regards with the page load sample size I've got samples varying from over 3,000 page views to 38,000. Do you think that is the smallest still a fair representation to be considered valid, please? I'd say so but not sure what is the threshold if you know what I mean.

 

Thanks and Best Regards, A.

 

 

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 4
Rising Star

Hi A,

 

That's difficult to answer.  The numbers you are quoting are most likley entirely adequate but what is the distribution?  Do they all occur on one or two days or are they more evenly distributed over your date range?  That said, you should use the data to drive testing; that is, the numbers should not be judge and jury on whether you have broswer specific issues, but point you to test the ones that seem to have issues.

 

Hope that helps!

 

Best,

 

Theo Bennett

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Average page load time by specific browsers not real average - how is it calculated?

ACG
Follower ✭ ✭ ✭
# 5
ACG
Follower ✭ ✭ ✭

It does Theo, thank you! 

Re: Average page load time by specific browsers not real average - how is it calculated?

ACG
Follower ✭ ✭ ✭
# 6
ACG
Follower ✭ ✭ ✭

Hi Theo,

 

 

I hope you don't mind me asking another question related to average page load time. 

 

I've been asked to check % of sessions that device types bring to a site month by month and the monthly average page load time. I created a custom report and exported it into Excel.

 

The monthly average page load time in my exported report which is calculated by averaging each month's desktop, mobile and tablet average page load time doesn't match with my generic average page load time on Analytics. 

 

I remember you said to not to average averages but if I need to understand what device type might have affected the average load time I need to check each device's load time but obviously this average doesn't match with the generic... 

 

Can you throw some light into why they don't match, please? I attach two screenshots at the end of this message. I can provide one screenshot that summarises everything if it's easier.

 

Thanks very much, A.

  

How average in exported report is calculated

How average in exported report is calculated

 differences_in_averagepageloadtime.JPGDifference from overall average and average from exported report

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 7
Rising Star

Hi A,

 

Ask away!  Smiley Happy  First things first though: With the volume you have here, you can set your sample rate to 100 percent (default is 1%).

 

Secondly, you can rely on GA's calculations in your custom report.   I would not use the data though unless you have an even distribution of samples for each type. 

 

GA is taking the number of samples for each and doing it's calculations.  You are then adding them up and dividing by 3 for your averages ( in essence).  What you don't know is how many samples are in each yet you've weighted them all equally.  If there is only on sample in desktop and 10000 in mobile and 500 in tablet GA will calculate the overall aerage by dividing by the 10501 and the cat average by dividing by 10000, 500 and by 1.  Hence why you should not average averages.  Smiley Happy

 

Best,

 

Theo

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 8
Rising Star

P.S.  If you want to take it a step further, you can use the Chrome Developer console to mimic mobile and tablet and the Audit tab to get recommendations on how to make your pages faster.   Smiley Happy

 

https://developer.chrome.com/devtools#audits

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Re: Average page load time by specific browsers not real average - how is it calculated?

ACG
Follower ✭ ✭ ✭
# 9
ACG
Follower ✭ ✭ ✭

Thanks very much for your tips Theo, really appreciated.

 

Best Wishes, A.

Average page load time by specific browsers not real average - how is it calculated?

Rising Star
# 10
Rising Star

Hi A,

 

No worries! 

 

Best,

 

Theo

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn